Some models of decision making assume humans have perfect information, but Dr. Erica Gralla knows that is never the case when working with complex systems. And since better models of decision making can dramatically improve system performance and outcomes, she is hard at work developing models to better reflect the structures and context of various systems in which decisions are made—including the (imperfect) information that is really available to decision-makers.
One system she studies is disaster response supply chains. After a hurricane strikes or an earthquake occurs, for example, responders at warehouses need to know what supplies the affected people need, where the cargo is most needed, and which routes are open. All of these are supply chain questions. Multiply these types of decisions over the many people and organizations that are part of a disaster response system, and optimal decision making quickly becomes very complicated.
According to Dr. Gralla, mega-retailers like Amazon and Walmart are very good at supply chain management, but their models assume a particular system structure that just does not fit in a disaster response situation.
“You couldn’t use Amazon’s model for disaster response,” she says. “For one thing, our goals are different. We’re not trying to minimize cost. In addition, there’s no single central authority directing all aspects of the response, so we can’t just use a model to find an optimal decision and expect it to be implemented by all responders.”
Instead of starting with those models, she tries to understand the context and structure of each system—in this case, the evolving goals and decentralized structure—and then figure out the right decision making approach based on that system. So she developed a decision support tool for disaster response transportation to help planners with just these sorts of supply chain challenges, one that works with imperfect information and limited decision authority.
“This problem sounds easy but it’s not that easy to make sure you’re actually doing the most good,” maintains Dr. Gralla. “These supply chain decisions add up to whether or not someone gets service, and this can mean prolonging suffering unnecessarily.”
Dr. Gralla studies other decision making problems, too. She is working currently on a project to design the right division of labor across teams working on spacecraft design, and on a project to support decisions about the best ways to leverage development aid to strengthen the Ugandan agricultural supply chain systems. As she explains, all of these problems relate to understanding the system’s context and structure and how to use the structure to enable people to make better decisions.
“Most of my field thinks about telling you what decision to make, and I don’t always think that that’s the right answer,” she contends. “I think about how we set up the information flow and coordination and leave the decision to a person who’s in a better position to understand things that a model will never understand, in some cases leaving room for intuition, in some cases leaving room for changing priorities.”